Welcome to AI Week 2025

Post Syndicated from Kenny Johnson original https://blog.cloudflare.com/welcome-to-ai-week-2025/

We are witnessing in real time as AI fundamentally changes how people work across every industry. Customer support agents can respond to ten times the tickets. Software engineers are reviewers of AI generated code instead of spending hours pounding out boiler plate code. Salespeople can get back to focusing on building relationships instead of tedious follow up and administration. 

This technology feels magical, and Cloudflare is committed to helping companies build world class AI-driven experiences for their employees and customers.

There is a but, however. Any time a brand new technology with such widespread appeal emerges, the technology often outpaces the tools in place to govern, secure and control the technology. We’re already starting to see stories of vibe coded apps leaking all their users’ details. LLM chats that were intended to only be shared between colleagues, are actually out on the web, being indexed by search engines for all the world to see. AI Agents are being given the keys to the application kingdom, enabling them to work autonomously across an organization — but without proper tracking and control. And then there’s the risk of a well-meaning employee uploading confidential company or customer data into an LLM, which then uses it to train future models.

Beyond internal data used for LLM training, content creators and media companies are also faced with a decision about how they want LLM scrapers and information retrieval bots to interact with their content. Cloudflare has found that it can be hundreds, or even thousands, of times harder to generate site traffic (and therefore ad revenue) from an AI response versus a search engine result.

We’re hearing more and more of these stories from CISOs, CIOs, Creators, and even CEOs. These leaders are faced with a difficult choice: clamping down on all AI usage and bots — or letting them run wild. There needs to be something in between. And for that to be a real option, the tools to manage and secure AI need to catch up to AI itself.

This week, that’s what Cloudflare is focused on. Welcome to AI Week! Over the coming week, we will focus on four core areas to help companies secure and deliver AI experiences safely and securely:

  • Securing AI environments and workflows: AI is incredibly powerful. The problem is, innovation is outpacing control — we want to change that. And as one of the few zero trust providers also building out AI infrastructure for the web, we’re uniquely positioned to be able to do so. 

  • Protecting original content from misuse by AI: AI Companies are devouring organic content as quickly as it’s created… and creators aren’t seeing any benefit. We want to give content creators control over the content that they have worked so hard to develop.

  • Helping developers build world-class, secure, AI experiences: the possibilities for developers to create new applications on top of (or even building with) AI are endless.  We want to allow developers to create AI driven applications that are as close to users as possible, with security controls built-in from day one.

  • Making Cloudflare better for you with AI: AI is changing the nature of interfaces. For example, finding and mitigating issues buried in thousands and millions of logs and events across website, employee, and email usage is something that used to be tedious — but now with AI, it can be made easy. We’re working day and night to integrate AI into Cloudflare itself to make things more efficient for ourselves and our customers.

Securing AI environments and workflows

As Artificial Intelligence innovation continues to accelerate at an unprecedented pace, the speed of its development is increasingly outpacing the implementation of robust security controls. This rapid advancement, while promising immense benefits, simultaneously introduces novel and complex security challenges that traditional measures are often ill-equipped to address. Organizations are finding themselves grappling with the inherent risks of adopting powerful AI tools without adequate safeguards, leading to vulnerabilities such as Shadow AI and the uncontrolled proliferation of AI models, making the development of specialized AI security paramount.

As we look around the zero trust space, none of the other providers are moving fast enough to keep up with AI’s pace of innovation. This is something we know a thing or two about — and after this week, if you’re worried about governing AI usage inside your organization, we will have you covered. 

We will be announcing new and powerful controls to detect Shadow AI and control unauthorized AI usage. Additionally, we’ve built options for teams to establish the “paved path” of AI tooling in an organization to supercharge employee productivity without sacrificing security. Finally, we’ll be announcing new ways of protecting your own models from poisoning or attacks.


Protecting original content from AI

The explosion of Large Language Models (LLMs) has also created a new challenge for content creators: the unauthorized scraping and training of their valuable content. Cloudflare recognizes the critical need for creators to maintain control over their intellectual property. That’s why we’ve introduced Crawl Control, a groundbreaking initiative designed to empower content owners to manage how their content is accessed and used by AI models.

In the past two months, we’ve seen incredible progress with Crawl Control. We’ve significantly expanded the number of participating content providers, allowing more creators to leverage this innovative protection. We’ve also refined our detection mechanisms to more accurately identify AI crawlers and ensure that only authorized access occurs. Furthermore, we’ve streamlined the integration process, making it easier for new publishers to onboard and begin protecting their content within minutes. Our goal remains to provide content creators with the tools they need to thrive in the age of AI, ensuring they are compensated and acknowledged for the content they produce.


Helping you build world-class, secure, AI experiences

We believe that AI experiences should have security controls by default. This is why we are heavily investing in both our developer platform’s AI Gateway and the associated security controls for those products. This two pronged approach allows developers to iterate and test new ideas without the fear of painful or embarrassing security issues.

The Cloudflare AI Gateway allows developers to deploy AI-driven applications with unparalleled speed and efficiency, ensuring that these applications are as close to end-users as possible. This proximity minimizes latency and maximizes performance, delivering a seamless and responsive user experience that is critical in today’s fast-paced digital landscape.

This week, we’re announcing significant enhancements to the AI Gateway, further solidifying its position as the premier platform for AI application deployment. These improvements include advanced caching mechanisms that reduce redundant model calls, leading to faster response times and lower operational costs. We are also introducing expanded observability features, providing developers with deeper insights into their AI model’s performance and usage patterns, which will enable more effective debugging and optimization. Furthermore, new integrations with popular AI frameworks and services will simplify the development workflow, allowing developers to leverage the AI Gateway’s benefits with even greater ease. Our commitment is to provide developers with the tools to innovate and deliver cutting-edge AI experiences to their users.

Making Cloudflare better with AI 

We’re integrating AI across our entire product suite to enhance the Cloudflare experience itself. From intelligent threat detection that adapts to emerging attack patterns, to AI-powered optimizations that fine-tune network performance, our goal is to leverage AI to make our platform more intuitive, efficient, and secure. We envision a future where Cloudflare’s products proactively anticipate user needs, automate complex tasks, and deliver unparalleled insights, all powered by seamlessly embedded AI. This commitment to internal AI integration ensures that as the digital landscape evolves, Cloudflare remains at the forefront of innovation, continuously delivering superior value to our users.

We cannot wait to share these updates and announcements with you. Follow our AI Week hub page for all the latest releases from our blog and CloudflareTV.


Позиция относно европейския регламент за т.нар. „чат контрол“

Post Syndicated from Bozho original https://blog.bozho.net/blog/4508

Интересът към една много чувствителна дигитална тема набира скорост в последните седмици – т.нар. „чат контрол“ – проект на регламент на ЕС, с който всяко съобщение, което изпращаме, дори с криптирани приложения, ще бъде сканирано за материали, съдържащи сексуална експлоатация на деца (т.нар. CSAM).

Ще направя дълга ретроспекция и обяснение на техническите проблеми, но преди това трябва да заявя, че позицията и на позицията и на Да, България, и на колегите в коалиция е, че не трябва да бъдат реализирани инвазивни мерки спрямо личната кореспондеция, които създават предпоставки за масовото ѝ следене, и съответно предложението и в оригиналния му вид, и във вида, в който датското председателство го вижда, е неприемливо.

Дори без текстовете за криптираните приложение, регламентът прави сериозни крачки към повишаване на ефективността на борбата с разпространението на CSAM, така че в предстоящо заседание на Съвета на ЕС през есента, горещият въпрос ще бъде именно криптираните приложения – по останалото по-скоро има консенсус, защото е безспорно, че трябва по-сериозно и ефективно противодействие на такива престъпления. Затова останалите текстове в регламента трябва да бъдат подкрепени.

Първоначално това предложение включваше възможност за изпращане на снимките централно към европейско звено за тяхното сканиране. Това беше посрещнато с бурно неодобрение, тъй като на практика елиминира криптирането от край до край – ако всяко съобщение, съдържащо снимка или линк бъде изпращано някъде, това на практика елиминира криптирането.

Затова при предходно председателство на Съвета на ЕС имаше работно предложение за ограничаване на тази мярка само до вече известно съдържание (CSAM) и сканирането на да се извършва само на устройството, преди криптиране, без да се изпраща никъде. Това на пръв поглед звучеше по-разумно, защото отдалечаваше предложението от масовото следене. Дори, на пръв поглед, изглеждаше, че може да се приложи и изкуствен интелект на самото устройство. Тогава направих такова допускане, с уговорката за внимателен анализ.

Само че, когато човек направи такъв внимателен анализ, става ясно, че и това е едновременно и опасно, и не особено полезно за постигане на целта. Ще изредя няколко детайла:

1. Организираните престъпни групи, които се занимават с разпространение на CSAM, просто ще започнат да използват свои приложения, които, благодарение на друг регламент на ЕС (DMA) ще могат да заредят в телефоните си, без те да отговарят на новите изисквания. Т.е. защитата на личната кореспонденция на обикновените хора ще бъде отслабена и ще бъдат създадени рискове за масово наблюдение и злоупотреби, а престъпните групи ще го заобикалят.

2. В момента няма технология, с която по работещ начин да се реализира желанието на датското председателство и на ЕК – алгоритмните за т.нар. perceptual hashing не са правени с цел защита от злонамерени модификации, т.е. с малко визуални ефекти и трансформации на снимките, те ще останат неразпознати. Също така, както тези алгоритми, така и моделите за изкуствен интелект, които биха работили на крайните устройства, дават фалшиво-позитивни резултати, което рискува наводняване на правоохранителните органи с напълно законни снимки. За да може да бъде въведена регулаторно такава технология, тя трябва да отговаря на всички тези (и други) предизвикателства – нямаме право затворени, експериментални технологии да бъдат част от нормативната уредба, още повече, когато се засягат основни конституционни права.

3. Технологията (ако евентуално някой ден бъде създадена достатъчно добра такава) трябва да е с отворен код, а ако ползва AI – да е с отворен модел и много ясен и прозрачен процес за одит на данните за трениране. Също така, централната база данни трябва да е обект на много строги процедури за подаване и проверка на съдържание, защото в противен случай държава членка с ниско ниво на върховенство на правото може да подава и друго съдържание, вкл. политическо такова, което иска да следи и цензурира. Примерът от миналото лято със свалянето на сатиричния сайт на „Ново начало“ е само индикация за това как може да се злоупотребява. Припомням, че тогава сайтът се появи в списъци на компании за киберсигурност като „съдържание за възрастни“ и беше блокирано в мрежи, където софтуер на тези компании беше инсталиран – тогава това вероятно беше направено от частни подизпълнители на Пеевски, но разлика в подхода няма.

Това са само част от аргументите защо предложението е недообмислено. Нужен е много по-дълъг дебат по темата и много повече научни статии, изследващи и развиващи технологичната готовност за такива подходи. Добрата новина е, че много държави все още се колебаят, а сред тях е Германия, и съответно няма мнозинство в Съвета, а мандатът на Европейския парламент е срещу такъв тип инвазивни промени.

Когато има легитимна критика към ЕС, тя е, че такъв тип регулации са възможни. Но отговорът на тази критика е, че явно държавите-членки държат на гаранциите за лична свобода и че в сериозен дебат в рамките на целия Европейски съюз могат да бъдат спирани оруеловските мерки и да бъдат намирани работещи решения вместо добре звучащи, но неработещи технологични регулации.

Материалът Позиция относно европейския регламент за т.нар. „чат контрол“ е публикуван за пръв път на БЛОГодаря.

XConn Tech Shows off New PCIe Gen6 and CXL 3 Switch Chips at FMS 2025

Post Syndicated from Cliff Robinson original https://www.servethehome.com/xconn-tech-shows-off-new-pcie-gen6-and-cxl-3-switch-chips-at-fms-2025/

At FMS 2025, we saw the new XConn Tech PCIe Gen6/ CXL 3 era switch chip running a live demo on the show floor

The post XConn Tech Shows off New PCIe Gen6 and CXL 3 Switch Chips at FMS 2025 appeared first on ServeTheHome.

FFmpeg 8.0 released

Post Syndicated from corbet original https://lwn.net/Articles/1034813/

Version 8.0 of the FFmpeg
audio and video toolkit has been released.

Thanks to several delays, and modernization of our entire infrastructure,
this release ended up being one of our largest releases to date. In short,
its new features are:

  • Native decoders: APV, ProRes RAW, RealVideo 6.0, Sanyo LD-ADPCM, G.728
  • VVC decoder improvements: IBC, ACT, Palette Mode
  • Vulkan compute-based codecs: FFv1 (encode and decode), ProRes RAW (decode only)
  • Hardware accelerated decoding: Vulkan VP9, VAAPI VVC, OpenHarmony H264/5
  • Hardware accelerated encoding: Vulkan AV1, OpenHarmony H264/5
  • Formats: MCC, G.728, Whip, APV
  • Filters: colordetect, pad_cuda, scale_d3d11, Whisper, and others

Amazon Redshift Serverless at 4 RPUs: High-value analytics at low cost

Post Syndicated from Ricardo Serafim original https://aws.amazon.com/blogs/big-data/amazon-redshift-serverless-at-4-rpus-high-value-analytics-at-low-cost/

Organizations across industries struggle with the economics of data analytics. High entry costs, complex capacity planning, and unpredictable workload demands create barriers that prevent teams from accessing the insights they need. Small businesses abandon analytics initiatives due to prohibitive minimums, and enterprises overprovision resources for development environments, leading to inefficient spending.

Amazon Redshift Serverless now addresses these challenges with 4 RPU configurations, helping you get started with a lower base capacity that runs scalable analytics workloads beginning at $1.50 per hour. This new option transforms the economics of data analytics with the flexibility to scale up automatically based on workload demands. You only pay for the compute capacity you consume, calculated on a per-second basis.

With 64 GB of memory and support for up to 32 TB of managed storage, this lower entry point offering addresses several common customer needs, including development and test environments that maintain separate workloads at lower cost and production workloads with variable demand that need cost-effective scaling. The configuration is particularly useful for test and development environments, departmental data warehouses, periodic reporting workloads, gaming analytics, and data mesh architectures with unpredictable usage patterns. Organizations just starting with cloud analytics can use this low-cost option while getting access to enterprise features like automatic scaling, built-in security, and seamless data lake integration.In this post, we examine how this new sizing option makes Redshift Serverless accessible to smaller organizations while providing enterprises with cost-effective environments for development, testing, and variable workloads.

New 4 RPU minimum base capacity in Redshift Serverless

Redshift Serverless measures compute capacity using Redshift Processing Units (RPUs), where each RPU provides 16 GB of memory. With this new minimum base capacity, the 4 RPU configuration delivers a total of 64 GB of memory. It supports up to 32 TB of managed storage, with a maximum of 100 columns per table. The 4 RPU configuration is cost-efficient, and it’s designed for lighter workloads. When your workload requires additional resources, Redshift Serverless automatically scales up the compute capacity. After you have scaled beyond 4 RPUs, your data warehouse will continue using the higher RPU level to maintain consistent performance. This behavior provides workload stability while preserving the benefits of automatic scaling.

For workloads requiring more resources, such as tables with a large number of columns or higher concurrency requirements, you can choose higher base capacities ranging from 8 RPUs up to 1024 RPUs. This flexibility helps you start small and adjust your resources as your analytics requirements evolve.

Benefits of Redshift Serverless with 4 RPUs

This new feature offers the following benefits:

  • Cost-effective entry point – The new 4 RPU configuration is a low-cost option for cloud data warehousing, making enterprise-grade analytics accessible to organizations of various sizes, such as startups exploring their first data warehouse or established enterprises optimizing their analytics spending. For example, in the US East (N. Virginia) Region, the compute cost is $0.375 per RPU-hour. For a 4 RPU base capacity, this translates to $1.50 per hour of active workload time. Because you’re only charged when workloads are running, small-scale users can keep costs predictable and low. This configuration helps teams begin their analytics journey with minimal upfront commitment. Development teams can maintain dedicated environments for testing and experimentation without significant cost overhead.
  • Support for smaller datasets – With support for up to 32 TB of Redshift Managed Storage, the 4 RPU configuration is well-suited for smaller data warehouses. It can handle datasets ranging from a few gigabytes to tens of terabytes, making it ideal for startups, small businesses, or departments with limited data volumes.
  • Seamless integration with the AWS ecosystem – The 4 RPU configuration integrates seamlessly with other AWS services, such as Amazon Simple Storage Service (Amazon S3) for data lakes, AWS Glue for ETL (extract, transform, and load), and Amazon QuickSight for visualization. This makes it straightforward to build end-to-end analytics pipelines, even for smaller-scale projects. Additionally, Redshift data lake queries on external Amazon S3 data are included in the RPU billing, simplifying cost management.
  • Use case flexibility – The 4 RPU configuration proves valuable across numerous analytics scenarios. Development and testing environments benefit from cost-effective isolation, and departmental data warehouses can start small and scale as needed. Organizations running periodic reporting workloads or proof-of-concept projects can optimize costs by paying only for actual usage. Even small to medium-sized production workloads can use this configuration effectively.

Regardless of the use case, you can benefit from the full feature set of Redshift Serverless, including built-in security, data lake integration, and automated maintenance.

Use cases for Redshift Serverless with 4 RPU workgroups

The 4 RPU configuration is tailored for scenarios where lightweight compute resources suffice. The following are some practical use cases:

  • Small business analytics – Small businesses with limited data (less than 32 GB) can analyze sales, customer behavior, or operational metrics with cost-effective data warehouses. Running 10–20 daily ETL queries and occasional one-time queries remains cost-effective at this capacity.
  • Development and testing environments – The configuration is well-suited for development and test environments where full production resources aren’t needed. Data engineers can experiment with Redshift Serverless, prototype queries, or build proof-of-concept solutions without committing to higher RPU capacities. The 4 RPU configuration lowers the cost of continuous integration and delivery (CI/CD) testing of data pipelines. Teams can run automated integration tests and schema validations in isolated environments that mirror production systems while optimizing costs through per-second billing.
  • Analytics for startups – Startups can build robust product analytics capabilities without significant upfront investment. Teams can track customer behavior, feature adoption, and KPIs using familiar SQL queries, then connect business intelligence (BI) tools like Quicksight or Tableau for lightweight dashboarding.
  • Training and experimentation – Organizations can create dedicated sandbox environments for data analysts’ onboarding and experimentation with minimal budget impact. These environments are perfect for exploring analytics powered by large language models (LLMs), semantic layer development, or generative AI applications.
  • Data quality workflows – The feature efficiently supports scheduled jobs for data quality validation, checking data freshness, integrity, and conformance without dedicating high-capacity environments to routine QA tasks.
  • Enterprise team enablement – Large organizations can implement decentralized data warehousing strategies. Each department can operate its data warehouse aligned with specific needs and budgets, enabling department-level chargeback models.
  • Environment isolation – Organizations can create dedicated workgroups per environment (development, test, QA, UAT), providing complete isolation without sharing compute resources or risking cross-environment interference.
  • Data mesh architecture – Domain teams can operate independently while maintaining cost-efficiency. Each domain runs its workgroup for lightweight transformations, domain-specific marts, and KPI calculations. It offers a flexible sizing option in a data mesh architecture.
  • Event-driven analytics – Well-suited for short-lived or event-triggered analytics tasks. Organizations can programmatically create workgroups through APIs for A/B test analysis, campaign performance summaries, or machine learning (ML) pipeline validation.
  • Low-volume one-time reporting – Organizations with infrequent or lightweight reporting needs, such as monthly financial summaries or dashboard refreshes, can use 4 RPUs to minimize costs while maintaining performance.

Cost considerations and best practices

Although the 4 RPU configuration is cost-effective, there are a few considerations to keep in mind to optimize expenses:

  • Billing – Redshift Serverless bills on a per-second basis with a 60-second minimum per query. For very short queries (such as subsecond), this can inflate costs. To mitigate this, batch queries where possible to maximize resource utilization within the 60-second window. For more information, see Amazon Redshift pricing.
  • Set usage limits – Use the Redshift Serverless console to set maximum RPU-hour limits (daily, weekly, or monthly) to prevent unexpected costs. You can configure alerts or automatically turn off queries when limits are reached. To learn more, see Setting usage limits, including setting RPU limits.
  • Monitor with system views – Query the SYS_SERVERLESS_USAGE system table to track RPU consumption and estimate query costs. For example, you can calculate daily costs by aggregating charged seconds and multiplying by the RPU rate.
  • Close transactions – Make sure transactions are explicitly closed (using COMMIT or ROLLBACK) to avoid idle sessions consuming RPUs, which can lead to unnecessary charges.

The following is a practical example for a 4 RPU workgroup in US East (N. Virginia) at $0.375/RPU-hour for a scenario of a 10-minute query running daily: This is compute costs only. Primary storage capacity is billed as Redshift Managed Storage (RMS).

  • Workload duration: 10 minutes (600 seconds)
  • Cost: (600 seconds / 3600 seconds) × 4 RPUs × $0.375 = $0.25
  • Monthly cost (30 days): $0.25 × 30 = $7.50

Performance considerations

Although the 4 RPU configuration is cost-efficient, it’s designed for lighter workloads. For complex queries or datasets exceeding 32 TB, you must set up 8 RPUs to 24 RPUs to support up to 128 TB of storage. For more than 128 TB, you need 32 RPUs or more. If query performance is a priority, consider increasing the base capacity or enabling AI-driven scaling and optimization to optimize resources dynamically. Benchmark tests suggest that higher RPUs (such as 32 RPUs) significantly improve performance for complex queries. However, for simpler tasks, 4 RPUs deliver adequate throughput.

To monitor performance, use the Redshift Serverless console or CloudWatch metrics like ComputeCapacity and ComputeSeconds. The SYS_QUERY_HISTORY table can also help analyze query runtimes and identify bottlenecks.

Conclusion

Redshift Serverless with 4 RPU represents a significant step forward in making enterprise-grade analytics cheaper and accessible to organizations of different sizes, such as a startup building its first analytics system, a development team looking to optimize testing environments, or an enterprise implementing a data mesh architecture. This new configuration combines the power and flexibility of Redshift Serverless with a cost-effective entry point, so teams can start small and scale seamlessly as their needs grow. The ability to begin with minimal commitment while maintaining access to advanced features like automatic scaling, built-in security, and seamless data lake integration makes this a compelling option for modern data analytics workloads. Combined with pay-per-second billing and intelligent resource management, Redshift Serverless with 4 RPU delivers the ideal balance of cost-efficiency and performance.

To get started with cost-effective analytics, visit the AWS Management Console to create your Redshift Serverless workgroup with 4 RPUs. For more information, refer to the Amazon Redshift Serverless Management Guide or Amazon Redshift best practices. Plan your analytics budget effectively using the AWS Pricing Calculator to estimate costs based on your specific workload patterns, or contact your AWS account team to discuss your particular use case.

About the authors

Ricardo Serafim

Ricardo Serafim

Ricardo is a Senior Analytics Specialist Solutions Architect at AWS. He has been helping companies with Data Warehouse solutions since 2007.

Ashish Agrawal

Ashish Agrawal

Ashish is a Principal Product Manager with Amazon Redshift, building cloud-based data warehouses and analytics cloud services. Ashish has over 25 years of experience in IT. Ashish has expertise in data warehouses, data lakes, and platform as a service. Ashish has been a speaker at worldwide technical conferences.

Andre Hass

Andre Hass

Andre is a Senior Technical Account Manager at AWS, specialized in AWS Data Analytics workloads. With more than 20 years of experience in databases and data analytics, he helps customers optimize their data solutions and navigate complex technical challenges. When not immersed in the world of data, Andre can be found pursuing his passion for outdoor adventures. He enjoys camping, hiking, and exploring new destinations with his family on weekends or whenever an opportunity arises.

The collective thoughts of the interwebz